The present invention relates generally to audio processing, and more specifically, to audio modification/substitution of a background noise generated during an event, e.g., during a sport event at a stadium.
Audio processing of a background noise during an event (e.g., a background noise during a sport event or during any performance in front of an audience) may represent a challenge. On one hand it is desirable to have a background noise during broadcasting of the event, e.g., along the voice of a commentator, to convey the reaction of an audience (e.g., applause, cheering), sounds related to the event itself (e.g., ball hit) and/or event procedural announcements or music, etc. This type of a background noise is desirable and is a “good” noise.
On the other hand, the challenge for the background noise processing is that some background noise sounds may be extremely loud causing an interference with the production audio (e.g., a commentator voice), which may be distracting to viewers of the event watching the event via television, computers or any other means. Such a loud noise is a “bad” noise which is highly undesirable and should be cancelled/reduced for eliminating its interfering with broadcasting.
The previous solutions utilizing standard noise cancellation using hardware and/or software solutions typically do not have an intelligence to distinguish between “good” and “bad” noises, e.g., a standard noise cancellation does not have an intelligence of discerning between different acoustic (audio) frequencies of the background noise. Also the standard noise cancellation solutions are relatively slow in order to achieve a desirable real-time performance.